- Регистрация
- 14.01.2009
- Сообщения
- 3 645
- Благодарностей
- 2 513
- Баллы
- 113
How to create a recognition module CAPTCHAs?
This is done in several steps:
(There are video below topic )
1) Create a new project. The project consists of recognized CAPTCHAs, pattern symbols (letters,digits, arithmetic operations, etc.) and trained recognition module. Do not forget to save projects if it was changed.
2) To add CAPTCHAs to the project you need to save 50-200 captchas into files. For each image CAPTCHA put near a text file with the same name as a CAPTCHA. This file have to contain a text written in a CAPTCHA. The easiest way to do it with the aid of a small template in ZennoPoster.
3) To add patterns in project, you need to cut them for each symbol in the template editor and send it to the project. If the CAPTCHA characters has a small slope, then for one character can be cut a few templates (an average of no more than 5 patterns on the symbol, otherwise recognition will run too slowly)
4) Run training. In the training are used recognized CAPTCHAs. Select the options for training in the dialogue that will start when you click "Training". Training time can be from 10 seconds till 10 hours. You can stop training and then continue or start over. You can continue training of any recognition module.
If more than 50% of the time learning passed and recognition is still 0%, it means you either have something wrong done, or this CAPTCHA can not be recognized by this version of the program.
5) Testing. Percentage of recognition, which you see in the training is a little bit higher than real percents of recognition. To specify the percentage of recognition and to test the speed of recognition, go to the tab "Testing module". Upload other (which are not used for training) recognized CAPTCHAs.
How to use module in ZennoPoster?
Put the saved project after the training of module in the folder /Modules of ZennoPoster. When you configure the recognition CAPTCHAs your project will be available in the drop-down list with the names of the modules of recognition.
Just select a module.
This is done in several steps:
(There are video below topic )
1) Create a new project. The project consists of recognized CAPTCHAs, pattern symbols (letters,digits, arithmetic operations, etc.) and trained recognition module. Do not forget to save projects if it was changed.
2) To add CAPTCHAs to the project you need to save 50-200 captchas into files. For each image CAPTCHA put near a text file with the same name as a CAPTCHA. This file have to contain a text written in a CAPTCHA. The easiest way to do it with the aid of a small template in ZennoPoster.
3) To add patterns in project, you need to cut them for each symbol in the template editor and send it to the project. If the CAPTCHA characters has a small slope, then for one character can be cut a few templates (an average of no more than 5 patterns on the symbol, otherwise recognition will run too slowly)
4) Run training. In the training are used recognized CAPTCHAs. Select the options for training in the dialogue that will start when you click "Training". Training time can be from 10 seconds till 10 hours. You can stop training and then continue or start over. You can continue training of any recognition module.
If more than 50% of the time learning passed and recognition is still 0%, it means you either have something wrong done, or this CAPTCHA can not be recognized by this version of the program.
5) Testing. Percentage of recognition, which you see in the training is a little bit higher than real percents of recognition. To specify the percentage of recognition and to test the speed of recognition, go to the tab "Testing module". Upload other (which are not used for training) recognized CAPTCHAs.
How to use module in ZennoPoster?
Put the saved project after the training of module in the folder /Modules of ZennoPoster. When you configure the recognition CAPTCHAs your project will be available in the drop-down list with the names of the modules of recognition.
Just select a module.
Последнее редактирование модератором: